Organizational Learning and Knowledge Management:Research Results To Date

Project Sponsor: numerous public and private sector organizations over the years

Barry G. Silverman, Principal Investigator
Univeristy of Pennsylvania, Philadelphia, PA
Barryg@seas.upenn.edu
April 2001

As complexity grows, organizations shift from top down command and control to decentralized management to rely on co-evolving, self-adaptive sub-groups and individuals. Organizations are inverting their structures to foster stakeholder collectives and emergence, and are only able to loosely couple the results. Current generations of software (e.g., customer relationship management, enterprise resource planning, workflow management, community chatting and knowledge-sharing websites, etc.) have many significant successes, but in other ways they still fail to fully support such communities -- in many respects they fail to fully integrate the data available; place added tasks on their users for mining and analyzing these impoverished datasets; use fixed, narrow rules and knowledge (business logic) that are brittle and bereft of situational understanding and wisdom; and rarely reduce the cognitive burden of the higher level mental tasks the users are attempting to undertake.   New, next generation approaches are needed to sustain, debias, and enhance the performance of these distributed, emergent parts so that the whole improves its capability. But what are these new approaches?

The goal of this research is to study human decision makers in complex and networked settings, to analyze performance obstacles and judgment biases,  to derive principles of design of software systems so they enhance rather than hinder human performance, and to develop and study new classes of agent technology that foster human abilities to shift their mindset and increase understanding and wisdom about their situation, tasks, and environments. This research straddles a number of projects on  modeling the human mindset/cognition in specific task-environments and to designing intelligent software and related workflow processes that attempt to shift and change behavior from the biased to the statistically debiased, from the individual muddler to the participative satisficer, from the narrow analytic with little corporate memory to the broad synthetic enabled with historical and contextual awareness. At various times these have taken the shape of decision rationale and corporate memory capture and analogical transfer systems, embedded critics/dialecticals, feedback systems/idea suggestors, virtual reality emotive coaches, embedded tutors, and natural language shopping agents. Often, the work involves designing a small network or  society of these agents, and then further studying human behavior in before/after field trials in actual enterprise settings (e.g., in the military, at NASA, in health care settings, etc.).

Some of my latest (still unfunded) ideas involve broad studies of the complex information environment we find ourselves in (multiple overlapping networks due to TV, radio, print, internet, person-to-person conversations, etc.) and how interactive message tailoring (agents) can affect decisionmaking. Related to this is the intriguing idea of enriching complexity-based game theoretic models to more accurately reflect the bounded rationality heuristics that community stakeholders utilize and to attempt to simulate and better understand human information processing and decisionmaking.

Selected efforts (see CV and publications for citations to these and a number of related papers):